EnQue

Product Support

Function

Pushes a tensor to a queue.

Prototype

• The source and destination locations do not need to be specified.

  1 2	template <typename T> __aicore__ inline bool EnQue(const LocalTensor<T>& tensor)

• The source and destination locations need to be specified.

  Bind VECIN and VECOUT by using TQueBind to implement VECIN and VECOUT buffer reuse. The following API is used to implement buffer reuse for vector computation. The source and destination locations need to be specified during enqueuing. In the scenario where vector computation is not involved, call bool EnQue(LocalTensor<T>& tensor).

  1 2	template <TPosition srcUserPos, TPosition dstUserPos, typename T> __aicore__ inline bool EnQue(const LocalTensor<T>& tensor)

Parameters

Restrictions

None

Returns

• true: The tensor is successfully added to the queue.
• false: The queue is full and the addition fails.

Example

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17

// API: EnQue Tensor
AscendC::TPipe pipe;
AscendC::TQue<AscendC::TPosition::VECOUT, 4> que;
int num = 4;
int len = 1024;
pipe.InitBuffer(que, num, len);
AscendC::LocalTensor<half> tensor1 = que.AllocTensor<half>();
que.EnQue(tensor1);// Add the tensor to the queue of VECOUT.
// API: EnQue specifies a specific src/dst position to join the corresponding queue.
// template <TPosition srcUserPos, TPosition dstUserPos> bool EnQue(LocalTensor<T>& tensor)
AscendC::TPipe pipe;
AscendC::TQueBind<AscendC::TPosition::VECIN, AscendC::TPosition::VECOUT, 1> que;
int num = 4;
int len = 1024;
pipe.InitBuffer(que, num, len);
AscendC::LocalTensor<half> tensor1 = que.AllocTensor<half>();
que.EnQue<AscendC::TPosition::GM, AscendC::TPosition::VECIN, half>(tensor1);// Add tensors to the VECIN queue to implement buffer reuse.